From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-0.8 required=3.0 tests=DKIM_SIGNED,DKIM_VALID, DKIM_VALID_AU,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,MAILING_LIST_MULTI,SPF_PASS autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id A9679C32789 for ; Mon, 5 Nov 2018 00:39:22 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 4F33120854 for ; Mon, 5 Nov 2018 00:39:22 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="ZBlf3C0v" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 4F33120854 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=gmail.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729822AbeKEJ4N (ORCPT ); Mon, 5 Nov 2018 04:56:13 -0500 Received: from mail-pl1-f196.google.com ([209.85.214.196]:41001 "EHLO mail-pl1-f196.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1725918AbeKEJ4N (ORCPT ); Mon, 5 Nov 2018 04:56:13 -0500 Received: by mail-pl1-f196.google.com with SMTP id p16-v6so3551239plr.8; Sun, 04 Nov 2018 16:39:18 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=subject:to:cc:references:from:message-id:date:user-agent :mime-version:in-reply-to:content-transfer-encoding:content-language; bh=6neK8wDxcPtfKxnxOpLM868G1Km25ksSSZAdztyqY+c=; b=ZBlf3C0vaM0MbxLHGFn/PcAdKfBNSlZmxHr+etBP6NynE8vdEnQO3OAOCao9d1NxnU v3NySkOjQv4Nu8EhierjlYmw9iuTI5AEqtxh8amytAne4Gc6+5RsP8S6z5EoHd1IYS/1 urLtgqHUoZ/QzgwZECIQSqQqTIIDHs6t3RzgL9oyHbx9A+eApBeEVyK3Qz9Ks4ltMkbA xc9Dud6ntRMY3yKlN8FodSD9IDB+P6pJp3v/491ldfwkc8MkTLErGmFPr46gMbwj48D5 5bw0gVLpcxl8QberWUkiG4MP+7YKkVVJMu80drZGpDKm8DohYAXVcJMj3TMqDxBNsf5S FQWw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:subject:to:cc:references:from:message-id:date :user-agent:mime-version:in-reply-to:content-transfer-encoding :content-language; bh=6neK8wDxcPtfKxnxOpLM868G1Km25ksSSZAdztyqY+c=; b=Ek49vkLmC1ZvuK3u2NJAEwutElw35t8gRL6/0hsDYnlCwYhMUryO6AvSfAu4ztLsOK 9gE6NIKOWv+tDwMfq8Io2ekcGuHRkftf0eyX8q+w9+zoFnUHaBayl/73HeKbxvUBid2D KdZ/9xabr/3hltEA3L9NrsrPYj8mar4wS9aAxw/dTxjwBJHI10ytR8B/c5GCKE2KGppn cSyHS/jL3otKKtfqzxn999mTpYGVVLzogPdcEttQw9UgQuWm+4O7cJRaQTQvNE1CvtSR 3fdtyk5jbHm3YHZgmE8WA4iPRXX1ZlZ9RfBx0nGiIBuBc8ymQVA7uzModmsc4ca767qZ XR1g== X-Gm-Message-State: AGRZ1gKa8r2nUB6JjQnUFtMrn2WbVN+prndrnsF/l7Mv64OX64wVbkkL X62IpgFeKiypQEhyWgYvwG/VV8J6 X-Google-Smtp-Source: AJdET5dPxn5yDHMpt9yMv0XfbYPl69Dh2F0Zrp17nxFuEIucGtDtO02E7n87i7hBKSpwpU0Pav7YDw== X-Received: by 2002:a17:902:a418:: with SMTP id p24-v6mr19344610plq.29.1541378357980; Sun, 04 Nov 2018 16:39:17 -0800 (PST) Received: from [0.0.0.0] (97.64.17.87.16clouds.com. [97.64.17.87]) by smtp.gmail.com with ESMTPSA id z1-v6sm23728282pgj.37.2018.11.04.16.39.13 (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Sun, 04 Nov 2018 16:39:16 -0800 (PST) Subject: Re: [PATCH v4 3/3] iio: magnetometer: Add driver support for PNI RM3100 To: Jonathan Cameron Cc: Jonathan Cameron , knaack.h@gmx.de, lars@metafoo.de, pmeerw@pmeerw.net, robh+dt@kernel.org, mark.rutland@arm.com, preid@electromag.com.au, himanshujha199640@gmail.com, linux-iio@vger.kernel.org, devicetree@vger.kernel.org, linux-kernel@vger.kernel.org References: <20181002143812.3661-1-songqiang1304521@gmail.com> <20181012073536.20339-1-songqiang1304521@gmail.com> <20181012073536.20339-4-songqiang1304521@gmail.com> <20181013111935.00a2e1af@archlinux> <20181021151427.5b3dbb9b@archlinux> <20181102092411.00003cb5@huawei.com> From: Song Qiang Message-ID: <71b9a137-4f7f-67c2-e83c-71f2c8cf9f75@gmail.com> Date: Mon, 5 Nov 2018 08:39:10 +0800 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.2.1 MIME-Version: 1.0 In-Reply-To: <20181102092411.00003cb5@huawei.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Content-Language: en-US Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2018/11/2 下午5:24, Jonathan Cameron wrote: > On Fri, 2 Nov 2018 15:55:27 +0800 > Song Qiang wrote: > >> On 2018/10/21 下午10:14, Jonathan Cameron wrote: >>> On Thu, 18 Oct 2018 16:24:15 +0800 >>> Song Qiang wrote: >>> >>> ... >>>>>> +static irqreturn_t rm3100_trigger_handler(int irq, void *p) >>>>>> +{ >>>>>> + struct iio_poll_func *pf = p; >>>>>> + struct iio_dev *indio_dev = pf->indio_dev; >>>>>> + unsigned long scan_mask = *indio_dev->active_scan_mask; >>>>>> + unsigned int mask_len = indio_dev->masklength; >>>>>> + struct rm3100_data *data = iio_priv(indio_dev); >>>>>> + struct regmap *regmap = data->regmap; >>>>>> + int ret, i, bit; >>>>>> + >>>>>> + mutex_lock(&data->lock); >>>>>> + switch (scan_mask) { >>>>>> + case BIT(0) | BIT(1) | BIT(2): >>>>>> + ret = regmap_bulk_read(regmap, RM3100_REG_MX2, data->buffer, 9); >>>>>> + mutex_unlock(&data->lock); >>>>>> + if (ret < 0) >>>>>> + goto done; >>>>>> + break; >>>>>> + case BIT(0) | BIT(1): >>>>>> + ret = regmap_bulk_read(regmap, RM3100_REG_MX2, data->buffer, 6); >>>>>> + mutex_unlock(&data->lock); >>>>>> + if (ret < 0) >>>>>> + goto done; >>>>>> + break; >>>>>> + case BIT(1) | BIT(2): >>>>>> + ret = regmap_bulk_read(regmap, RM3100_REG_MY2, data->buffer, 6); >>>>>> + mutex_unlock(&data->lock); >>>>>> + if (ret < 0) >>>>>> + goto done; >>>>>> + break; >>>>> What about BIT(0) | BIT(2)? >>>>> >>>>> Now you can do it like you have here and on that one corner case let the iio core >>>>> demux code take care of it, but then you will need to provide available_scan_masks >>>>> so the core knows it needs to handle this case. >>>>> >>>> This confused me a little. The available_scan_masks I was using is {BIT(0) | >>>> BIT(1) | BIT(2), 0x0}. Apparently in this version of patch I would like it to >>>> handle every circumstances like BIT(0), BIT(0) | BIT(2), BIT(1) | BIT(2), etc. >>>> Since Phil mentioned he would like this to reduce bus usage as much as we can >>>> and I want it, too, I think these three circumstances can be read consecutively >>>> while others can be read one axis at a time. So I plan to let  BIT(0) | BIT(2) >>>> fall into the 'default' section, which reads axis one by one. >>>> >>>> My question is, since this handles every possible combination, do I still need >>>> to list every available scan masks in available_scan_masks? >>> Ah. I see, I'd missed that the default was picking up that case as well as the >>> single axes. It would be interesting to sanity check if it is quicker on >>> a 'typical' platform to do the all axis read for the BIT(0) | BIT(2) case >>> and drop the middle value (which would be done using available scan_masks) >>> or to just do two independent reads. >>> >>> (I would guess it is worth reading the 'dead' axis). >>> >>>> All other problems will be fixed in the next patch. >>>> >>>> yours, >>>> >>>> Song Qiang >>>> >>>> >>>> ... >>> Thanks, >>> >>> Jonathan >> I tested this two ways of getting data with the following code snippet: >> >> >>     u8 buffer[9]; >>     struct timeval timebefore, timeafter; >> >>     do_gettimeofday(&timebefore); >>     ret = regmap_bulk_read(regmap, RM3100_REG_MX2, buffer, 9); >>     if (ret < 0) >>         goto unlock_return; >>     do_gettimeofday(&timeafter); >>     printk(KERN_INFO "read with dead axis time: %ld", >>            timeafter.tv_sec * 1000000 + timeafter.tv_usec - >>            timebefore.tv_sec * 1000000 - timebefore.tv_usec); >>     do_gettimeofday(&timebefore); >> >>     ret = regmap_bulk_read(regmap, RM3100_REG_MX2, buffer, 3); >>     if (ret < 0) >>         goto unlock_return; >>     ret = regmap_bulk_read(regmap, RM3100_REG_MZ2, buffer + 6, 3); >>     if (ret < 0) >>         goto unlock_return; >>     do_gettimeofday(&timeafter); >>     printk(KERN_INFO "read two single axis time: %ld", >>            timeafter.tv_sec * 1000000 + timeafter.tv_usec - >>            timebefore.tv_sec * 1000000 - timebefore.tv_usec); >> >> >> And get this result: >> >> [  161.264777] read with dead axis time: 883 >> [  161.270621] read two single axis time: 1359 >> [  161.575134] read with dead axis time: 852 >> [  161.580973] read two single axis time: 1356 >> [  161.895704] read with dead axis time: 854 >> [  161.903744] read two single axis time: 3540 >> [  162.223600] read with dead axis time: 853 >> [  162.229451] read two single axis time: 1363 >> [  162.591227] read with dead axis time: 850 >> [  162.597630] read two single axis time: 1555 >> [  162.920102] read with dead axis time: 852 >> [  162.926467] read two single axis time: 1534 >> [  163.303121] read with dead axis time: 881 >> [  163.308997] read two single axis time: 1390 >> [  163.711004] read with dead axis time: 861 >> >> >> It seems like you're right! Reading consecutively 9 bytes does save a lot time >> compared to read 3 bytes twice. >> > I've done this stuff before ;) We had this on the adis16365 parts back > in the early days of IIO. A worse case as it has a lot more channels, > but otherwise similar from what I recall. > > It would be an interesting exercise to trace those paths and find out the > balance between real hardware stuff we can't change and potential software > overheads. > > Chances are this is mostly 'real' stuff though but would be great to > confirm this. It's been on my list of things to do for years (not on > this driver obviously but in general)... > > Jonathan > I think I can try to use ftrace to trace it's flow path on my platform. I don't familiar with it, may need some time. yours, Song Qiang